Reflector Emissivity Biases on ATMS

- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner
Wednesday, 7 January 2015: 5:15 PM
230 (Phoenix Convention Center - West and North Buildings)
Kent L. Anderson, Northrop Grumman Electronic Systems, Azusa, CA; and V. Jacobo
Manuscript (409.1 kB)

1. INTRODUCTION The Suomi NPP spacecraft pitch-over maneuver revealed that the Advanced Technology Microwave Sounder (ATMS) [1,2] had a scan angle-dependent bias when viewing deep space, which could introduce errors up to 0.5 K in derived operational scene brightness temperatures. It was hypothesized that this bias was due to the polarized emissivity of the scanning reflector. For the ATMS configuration, the emissivity is polarization dependent, resulting in a scan-dependent bias, with the quasi-vertical channels having a different functional dependence than the quasi-horizontal channels. The normal-incidence emissivity of the scanning reflector was empirically estimated by fitting the theoretical model to the pitch-over maneuver data. Similarly, ground calibration data were used to estimate reflector emissivity of both the Suomi NPP and the JPSS-1 ATMS units. A special test was then performed on a spare reflector to make a direct measurement of its polarized emissivity. This paper presents the theoretical model, describes the special reflector emissivity test, and compares the results from the pitch maneuver, the ground calibrations and the special spare reflector test. A new calibration algorithm is then proposed as a means for correcting this scan-dependent bias.


The ATMS scanning reflector is a gold-plated beryllium flat plate, oriented 45 degrees relative to the wavefront. For a thin the conductive layer, the emissivity can be expected to significantly exceed the theoretical (Hagen-Rubens) emissivity of a perfectly flat bulk material, and can only be determined experimentally. Based on the Fresnel equations for reflections from a plane interface, the emissivity for polarization in the plane of incidence will be twice the polarization component normal to the plane of incidence. When the reflector is scanned relative to a fixed linear polarization feed horn, the Quasi-Vertical (QV) and Quasi-Horizontal (QH) components of emissions therefore have a sin2 and cos2 scan angle dependence, respectively. The resulting antenna temperature, viewed by the feed horn, also exhibits this sinusoidal scan-dependency, with a magnitude proportional to the difference between the reflector physical temperature and the scene brightness temperature.


During the Suomi NPP pitch-over maneuver, brightness temperature data were collected while executing the normal operational scanning profile, which consists of 96 samples in an “Earth-view” sector that covers a ± 52.7° swath, plus a 4.4° cold-space view calibration sector and a 4.4° internal warm load calibration sector. For the pitch-over orientation, the “Earth-view” swath was centered at the cold-space zenith orientation. The resulting data showed a very good fit to the sin2 and cos2 functions expected for the reflector polarized emissivity effect. There are, however, two kinds of small deviations from the sinusoidal functions: 1) higher than expected brightness temperature at the “cold-calibration” sector, and 2) asymmetry of the sun-side versus the anti-sun-side of the “Earth-view” sector. The former is explainable as a contribution from Earth intercept of the antenna sidelobes, and the latter as a contribution from spacecraft intercept. Therefore, to obtain a more accurate characterization of the reflector emissivity, the sin2 and cos2 curve fitting was applied only to the anti-sun half of the “Earth-view” sector. Based on an estimated reflector physical temperature, a value of normalized emissivity was derived by minimizing mean-squared errors of computed brightness temperature. The resulting emissivities ranged from 0.002 to 0.004 for the 22 channels.


In thermal vacuum ground calibration, the ATMS executes the normal operational scanning profile, observing a cold target at the cold-calibration sector and a variable scene-target for a portion of the Earth-view sector. When the scene target is at the same physical temperature as one of the calibration-view targets, any difference between the inferred scene brightness temperature and the calibration temperature is generally attributed to relative bias errors in the targets. Such an offset is expected to be nearly constant across all the channels covered by a set of targets. However, since the scene and calibration target views are at different scan angles, there will also be an error contribution due to the polarized emissivity of the scanning reflector. In particular, this error component will be significantly different for the QV and the QH channels. This effect has been consistently observed in the ground calibration data, and it is therefore possible to use this data to estimate the reflector emissivity. The results of this analysis for the Suomi NPP instrument calibration are consistent with the observations from the pitch-over maneuver. Estimates of reflector emissivity are also presented for the JPSS-1 reflector, based on it's ground calibration test.


A special test was performed using an ATMS spare flat-plate reflector and engineering models of the ATMS radiometric receivers and feedhorns. The measurements were conducted at five frequency bands: the ATMS K, Ka, V, W and G bands. The reflector was rotated by a servo drive motor and the radiation was measured by a linear-polarized feed horn and receiver assembly. This test produced the expected sinusoidal scan angle variations, at a frequency of twice the rotation rate. The test results show conclusively that the reflector emissivity effect is present for the flight spare reflector, with magnitudes in reasonable agreement with the estimates for the Suomi NPP flight unit, and with the same general relative magnitudes between the frequency bands.


A new radiometric calibration algorithm is presented, which will correct for this scan-angle dependent polarized emissivity effect. The algorithm is a function of scene temperature and reflector physical temperature, and corrects for biases introduced in the calibration sectors as well as for the scene sector. An error analysis is also reported, indicating the sensitivity of residual errors to the errors in the estimates for emissivity and reflector temperature.